Advances in the Use of Patient Reported Outcome Measures in. Electronic Health Records

Transcription

1 Advances in the Use of Patient Reported Outcome Measures in Electronic Health Records Including Case Studies November 7, 2013 In support of the PCORI National Workshop to Advance the Use of PRO measures in Electronic Health Records Atlanta, GA. November 19-20, 2013 Contact Information: Albert W. Wu, MD, MPH Center for Health Services and Outcomes Research Johns Hopkins Bloomberg School of Public Health 624 North Broadway Baltimore, Maryland

2 This technical report was written by Albert W. Wu, MD, MPH Johns Hopkins Bloomberg School of Public Health, Baltimore, MD Roxanne E. Jensen, PhD Lombardi Comprehensive Cancer Center Georgetown University, Washington, DC Claudia Salzberg, MS Johns Hopkins Bloomberg School of Public Health, Baltimore, MD Claire Snyder, PhD Johns Hopkins University School of Medicine, Baltimore, MD 1

3 TABLE OF CONTENTS Executive Summary Introduction... 7 What is a Patient Reported Outcome? RO 9 Taxonomy of PROs.. 9 Model for use of PROs for care, quality and research Electronic Health Records.. 14 Methods. 16 Case Studies Synthesizing Across the Case Studies 41 Summary of Systems.. 41 Consideration of Cases by System Features.. 42 Major Themes Moving Ahead: Standardization. 50 Conceptual System Architecture 51 Unanswered Questions Key Barriers and Enabling Factors 58 Conclusion. 64 References 65 Figures Acknowledgments. 81 2

4 EXECUTIVE SUMMARY The goal of health care systems is to obtain optimal patient outcomes, decrease risk and disease, and improve or maintain functioning for individual and populations. Incorporating the patient perspective through patient reported outcome (PRO) measures is a crucial element for clinical care, quality performance management and clinical research. PROs are any report coming directly from patients regarding their health condition and treatment, including symptoms, functional status and health-related quality of life. Some PRO measures are generic and appropriate for use in a wide range of conditions, while others focus on the specific symptoms and side effects of a given disease, condition or treatment. The use of PROs as outcome measures in research studies dates back to the 1980s. Since then, PRO data collection has increasingly integrated into health care. There is now a convergence in the evolution of PRO measurement, medical record keeping and comparative effectiveness research into an increasingly electronic and patient-centered space. Electronic health records (EHRs) began as an electronic version of the patient record for hospitals and clinics, and have evolved to serve a broader purpose of giving multiple stakeholders, including providers, managers and patients, access to a patient s medical information across different facilities. Systems have been developed recently that link EHRs to the collection of PRO data. One advantage of this linkage is that data collected for one purpose can potentially be used for multiple different tasks, including clinical care, quality assessment and improvement, research, and public reporting. Pilot studies, implementation efforts integrating PROs into EHRs and development of PRO research methods have received major federal support from the National Institutes of Health (NIH), Agency for Healthcare Research and Quality (AHRQ), Patient Centered Outcomes Research Institute (PCORI), Centers for Medicare and Medicaid Services (CMS), Food and Drug Administration, and the Office of the National Coordinator for Health Information Technology (ONC), as well as private and professional organizations. PCORI has organized a National Workshop to Advance the Use of PRO Measures in Electronic Health Records (EHRs), to be held on November 19-20, 2013 in Atlanta, GA. This paper provides a landscape review of the current state of use of PRO measures in EHRs, focusing on themes in the implementation and integrations of PROs within EHRs. 3

5 We focused on PRO data collection systems that were linked to an EHR system. We did not include efforts to measure patient satisfaction independent of patient health status, electronic reviews of systems, family history, health behaviors or health care utilization. The review includes 11 case studies from the US that illustrate the range of what has been done in diverse health care settings for clinical care, quality improvement and research. To supplement our review, we interviewed the principle developers and users of the systems. Many of the cases used the Epic Corporation EHR with its MyChart tethered patient portal for PRO data collection. The systems were based at Dartmouth- Hitchcock Medical Center, Cleveland Clinic, Group Health Cooperative, Cincinnati Children s Hospital, Kaiser Permanente Colorado (KPCO), Essentia Health/Minnesota Community Measurement, University of Pittsburgh Medical Center (UPMC), Duke University Medical Center, the University of California Los Angeles Medical Center (UCLA) and the University of Michigan Medical Center, and the University of Washington. The systems represent a broad range of functionality, patient populations and applications. Efforts at three health care plans (KPCO, Group Health and Essentia Health) illustrate applications of PRO collection at the plan level for clinical care, population based screening and quality of care evaluation. Hospital based efforts at Cincinnati Children s Hospital, the Cleveland Clinic and Dartmouth show how the use of PROs in specialty care can expand within a hospital. The review also includes examples of clinic-based, disease specific PRO collection (UCLA/Michigan s My GI Health, University of Washington Center for AIDS Research Networks of Clinical Systems (CNICS), Duke s Patient Care Monitor for cancer, and UPMC for primary care), some of which show how efforts may spread to specialists outside of the originating institution. While clinical care efforts focus primarily on providing the physician information to use during a patient visit, some systems also elicit information for followup evaluations. Essentia, through Minnesota Community Measurement, reports their scores as part of a statewide public reporting effort. A few organizations have integrated their systems into clinical and comparative effectiveness research (UCLA/Michigan, Cleveland Clinic, Cincinnati Children s, Duke), while others, like UPMC s system was designed to focus exclusively on clinical utility. 4

6 System features were examined across the 5 following categories: system design and implementation, measure selection, administration and data collection, reporting and interpretation, and analysis. While some of these system features are upstream from the technical aspects of EHR integration, each element builds the foundation for the focus, validity, interpretation and usefulness of the PRO data available in an EHR system. Overall the largest variations across these features were seen between systems that were designed using EHR-based PRO collection (e.g., Epic s MyChart feature) and outside collection that then sought to integrate information after collection and reporting. Systems that designed and collected PROs independently from the EHR, presented a much different approach allowing greater freedom in PRO content selection, more flexibility in patient access and score use, and limited ability to integrate PRO data with other clinical care markers. These trade-offs have implications at the person-, provider- and national-level which can be guided at this early point through standardization, developing a broad conceptual system architecture to encourage PRO collection compatible with larger research and evaluation efforts. In our analysis, four major themes emerged regarding the integration and use of PROs in EHRs: (1) Necessity of System Customization, (2) Balancing Research and Practice Goals, (3) Demonstrating Value and (4) EHR Integration and Limitations. These themes were important considerations for all case studies leading to key decisions ranging from design (e.g., PRO content and selection) to intended use (e.g., clinical care, research or quality improvement). Systems did not necessarily make similar choices. For example the University of Washington chose to scale-up PRO collection to implement standardized PRO collection in a national research data network infrastructure, while Kaiser Permanente Colorado s effort has focused on patient screening and quality improvement. However each of these systems has considered these themes with respect to the feasibility to sustain and expanding PRO collection efforts into other patient populations and/or clinical settings. There are a number of knowledge gaps identified in this report. These gaps center on the optimal system design features for PRO collection and integration. While these span PRO selection, administration analysis and security they all center around two main questions: the accuracy and accessibility of PRO data in an EHR. Ultimately these 5

7 current gaps point to the necessity of multidisciplinary teams and identifying teachable moments to educate clinicians, staff, researchers, and patients on PRO use. There are a number of remaining barriers to sustainable PRO integration. These fall under three broad categories: patient, clinician, and system functionality. The most common barriers were related hidden elements linked to electronic PRO collection that are found even in the most developed systems. Regardless of the number of system features and staff expertise, system awareness, response rate, clinician use, and consistent system access (e.g., enough tablets available in clinic, Wi-Fi access) all rely on engagement from patients, clinicians, and staff. Fortunately, most barriers identified have been shown to be somewhat modifiable, with enablers that may be scalable. To date, the perceived benefits of using PROs in clinical care have driven the implementation of in-clinic PRO data collection and EHR integration. Recent efforts to support PRO collection through patient portals offer a platform to further coordinate and develop PRO collection beyond the clinical encounter, further enhancing patient PRO monitoring for clinic, research and QI purposes. However, other options for collecting PROs are still necessary, including interactive voice response and in-clinic reporting. There is a diversity of approaches to PRO integration, and coordinated efforts are needed to increase the capacity to use them within EHRs for comparative effectiveness research. Barriers at the level of the patient, clinician and health system seem to be modifiable. There are considerable knowledge gaps regarding many scientific and practical aspects of implementing PRO measures into EHRs. Funding agencies and government bodies can support targeted research, infrastructure recommendations, education recommendations and methods development to help overcome current barriers, and ensure PROs can support the delivery high quality, patient centered care. 6

8 INTRODUCTION The goal of health care systems is to obtain optimal patient health outcomes, decrease risk and disease, and improve or maintain functioning for individual people and populations, by efficiently delivering services of the highest quality that is, services that are safe, timely, equitable, effective, efficient and patient centered (IOM 2001). Health care systems should learn from experience, by collecting and converting data about care and outcomes into knowledge. And such knowledge should be implemented into evidence-based clinical practice, driving improvements and the process of discovery as a natural outgrowth of patient care (IOM 2012). Health care choices made in collaboration between individual patients and their providers are central to this process. Incorporating the patient voice and perspective through patient reported outcomes (PRO) measures is critical for clinical care, quality performance management, and clinical research. In recent years there has been a convergence of trends in the measurement of health, the evolution of medical records, and the development of comparative effectiveness research (Wu 2013). Figure 1 depicts the evolution and convergence these trends. The vertical axis indicates increasing patient-centeredness, and the horizontal axis indicates increased digitization. As the measurement of health outcomes has come to more consistently include the patient s own assessment of his or her overall health and wellbeing, the science of PRO assessment has advanced, and the electronic collection and storage of health data has become routine. Paper-based medical records have been converted into electronic health records (EHRs), which can include customizable, builtin or tethered patient portals. Comparative effectiveness research has become more patient centered, with increased emphasis on stakeholder participation and capturing the patient perspective on treatments and outcomes. Consequently, PROs, EHRs, and comparative effectiveness research have converged in an increasingly patient-centered and digital space, providing the opportunity for the routine implementation of clinical systems to collection patient-reported information. PRO data collection is increasingly being integrated into health care. In the US, the National Institutes of Health, Agency for Healthcare Research and Quality, and Patient Centered Outcomes Research Institute (PCORI) have supported the development of 7

9 PRO methods for use in research and clinical practice (Lauer 2010, Wu 2010, Selby 2012). The Centers for Medicare and Medicaid Services and other payers, as well as the Food and Drug Administration, use PROs to evaluate interventions and programs(fda 2009). The Office of the National Coordinator for Health Information Technology (ONC) has supported the use of PROs by allowing their use as evidence that providers are making meaningful use of EHRs to improve quality of care or patient centeredness, and are therefore eligible for incentive payments (What is Meaningful Use 2013). In 2012, the National Quality Forum, the American Medical Associationconvened Physician Consortium for Performance Improvement (PCPI) and the American Society for Clinical Oncology all began initiatives to support the use of PRO measures for quality measurement and improvement (NQF 2012a,b; personal communication, from Kristen McNiff, October 24, 2013). In September 2013, the Institute of Medicine formed a Committee for Social & Behavioral Domains in Electronic Health Records (The National Academies 2013) chaired by the Director of the NIH Office of Behavioral and Social Science Research and PRO expert, Robert Kaplan. PCORI is organizing a National Workshop to Advance the Use of PRO measures in Electronic Health Records to be held on November 19-20, 2013 in Atlanta, Georgia. The workshop aims to review the current state of use of PROs in EHRs, identify barriers and facilitators to incorporating PRO measures in EHRs and identify specific actions PCORI and other organizations can take to support and promote the expanded use of PROs in EHRs. In support of the meeting, this paper provides a landscape review (i.e., non-systematic review) of the current state of use of PRO measures in EHRs. The pragmatic, rather than comprehensive, nature of the review is necessary, as surprisingly little has been published to date on most of the leading systems to measure PROs in EHRs. The review includes descriptions of a broad range of initiatives currently underway across the health care system to integrate PROs with the EHR. The 11 case studies illustrate the feasibility of integrating PRO measurement systems in various clinical, health plan, and population-based settings, and the utility of using PROs across clinical care, quality improvement, and research settings. This review also discusses features of system design and highlights key elements central to integrating PROs in EHRs on a larger, 8

10 broader scale including barriers, enabling factors and current knowledge gaps. What is a Patient Reported Outcome (PRO)? With the recent emphasis on patient-centered care and research (Selby 2012; IOM 2001, 2012), there is also increasing awareness of the importance of incorporating the patient s perspective in quality measurement and improvement. One approach for systematically capturing the patient s perspective is the routine collection of patientreported outcomes (PROs). PROs are defined as any report coming directly from patients about their health condition and treatment (FDA 2009) and include a range of outcomes such as symptoms, functional status, and health-related quality-of-life (Acquadro 2003). Some PRO measures are generic and appropriate for use in a wide range of diseases, as well as healthy populations; other PRO measures focus on the symptoms and side effects of a given disease, condition or treatment (Patrick 1989). There is a long history of using PROs as outcome measures in research studies dating back to the 1980s (Tarlov 1989; Katz 1987; Lohr 1987; Lohr 1989; Lohr 1992; Lipscomb 2005; Brundage 2007; Bottomley 2007) and somewhat more recently, examples and investigations of using PROs for individual patient care (Nelson 1990; Wasson 1999; Meyer 1994; Greenhalgh 2009; Snyder 2009a; Greenhalgh 1999; Valderas 2008; Marshall 2006; Greenhalgh 2005; Aaronson 2011). A real advantage of PRO assessment is that the data collected for one purpose can be used in multiple different ways including clinical care, quality assessment, quality improvement, research and public reporting (Wu 2013). Taxonomy of PROs Greenhalgh (2009) proposed a taxonomy for the different applications of PROs in clinical practice. This taxonomy classifies whether the PRO data are used at the individual or aggregated-level and whether the PRO data are used directly or indirectly to inform patient care. For example, when an individual patient completes a PRO questionnaire and that patient s data are provided to his/her provider(s), the data can be used at the individual level to screen for clinical problems, monitor progress over time, or promote patient-clinician communication. If this information is aggregated across a group of patients (e.g., at the provider or clinic level, or for a subgroup of patients) this can be used to inform quality improvement or conduct population monitoring. 9

11 Model for Use of PROs for Clinical Care, Quality Improvement and Research Complementing the Greenhalgh taxonomy, Snyder and Wu (Snyder 2013a) have proposed a model that describes the cycle of the use of PROs for quality assessment and improvement (Figure 2). This model demonstrates how the different aspects of the Greenhalgh taxonomy relate to each other, and can be used in a streamlined approach. For both the Greenhalgh taxonomy and the Snyder & Wu model, it is possible to implement either the full spectrum of applications, or one or more selected applications. Thus, there are a wide range of opportunities for using PROs in quality measurement and improvement. For all applications, the cycle begins with assessing the PROs (Box 1). The assessments may come from a number of sources, including clinical practice applications, research studies and population surveys, as described in detail elsewhere (Snyder 2013b). When PROs are used for clinical practice, they are collected from a patient with the intention of informing his/her care and management. To date, the large majority of integration with EHRs has been done for this purpose. When captured in research settings, the most common use of PROs, it is usually as outcome measures in clinical trials and observational studies and often occurs outside of EHR systems. In this application, the primary purpose of the PROs is to describe the impact of various diseases and/or treatments on measures of health extending beyond clinical endpoints. Finally, PROs can be collected as part of population-based surveys that provide a patient-centered perspective to complement other statistics (e.g., mortality rates)(barr 2003). In many cases, there is the potential to use these PRO data, regardless of the original purpose for their collection, to evaluate the quality of care (Box 3). An example of this is the United Kingdom s National Health Service (NHS) Patient-Reported Outcome Measures (PROMS) initiative ( (NHS 2013). Specifically, the NHS is evaluating the quality of care for select surgical interventions, including hip replacement, knee replacement, varicose vein procedures, and groin hernias. Patients complete pre-procedure and post-procedure PROs, and these data provide insight into the patient-centered value of the procedures overall. For the period from April 2012 to March 2013, there were nearly 240,000 procedures for 10

12 which PRO surveys were to be collected. For the approximately 163,000 preprocedures surveys returned, there were 88,000 post-operative questionnaires returned (NHS 2013). These surveys provide valuable insights into the impact of these surgical procedures on patient s functioning and well-being. For example, the proportion of patients reporting improvements on a general health status measure (the EQ-5D Index) ranged from 50% for groin hernia respondents to 87% for hip replacement respondents. The gains on the disease-specific measures were even greater, with 83% of varicose vein, 92% of knee replacement, and 95% of hip replacement patients reporting improvement. In addition to providing overall perspectives on the impact of these surgical procedures, there is also the opportunity to compare different hospital providers, though this may require case-mix adjustment (Devlin 2010). While the NHS example is one of the largest applications of PROs for quality measurement, other groups are also using, or planning to use, PROs in this way. For example, the American Society of Clinical Oncology is exploring the incorporation of PRO measures as quality indicators in its Quality Oncology Practice Initiative, focusing first on some common symptoms such as pain and nausea (personal communication, from Kristen McNiff, October 24, 2013). PROs may also be used for population health screening. In this application, PRO measures may be administered for disease screening, or for health risk assessment Nelson 2012). Disease screening programs can be used to identify untreated disease, such as depression in clinical care (American College of Surgeons 2011) or more broadly in the general population. Screening for health risks, for example, can be used to engage and motivate individuals to pursue changes in health behaviors, and the selfmanagement of chronic conditions (Shekelle 2003). Health risk assessment can also be effective at inspiring the uptake of prevention and health promotion activities in employee health programs. Linking health promotion to health care visits and physician advice can potentiate the benefits of PRO screening. While it is possible to go directly from PRO assessment (Box 1) to quality measurement (Box 3), an alternative approach would also use the data to improve the quality of individual patient management, as well (Box 2). The use of PROs in clinical practice involves having patients not only complete the questionnaires but making an 11

13 individual patient s assessment available to the patient s provider(s) to inform that patient s care. Multiple systems have been developed for collecting PROs and using them for clinical practice (Jensen In Press, Bennett 2012, Rose 2009, Basch 2009). For example, at Johns Hopkins, we have developed the PatientViewpoint webtool that was linked to the institution s home-grown electronic medical record ( (Snyder 2009, Snyder 2012, Hughes 2012, This tool enables clinicians to order PRO questionnaires much in the same way that they order lab tests or imaging studies. Patients receive an when it is time to complete a questionnaire and the results are provided to both the patient and clinician. The use of PROs in clinical practice can improve patient-clinician communication, and can also have an effect on patient care and outcomes (Hayward 2006, Valderas 2008, Marshall 2006, Greenhalgh 1999, Greenhalgh 2005, Velikova 2004, Velikova 2010, Berry 2011, Santana 2010, Detmar 2002, Bliven 2001, Boyce 2013, Espallargues 2000, Gutteling 2008, Lyndon 2011, Taenzer 2000, Takeuchi 2011) Thus, the PRO collection in itself can be an intervention with the intention of improving individual-level patient care. As noted above, using PRO data for individual patient care in no way precludes using the data for quality measurement. In fact, it facilitates the process if collected in a systematic way. It is feasible to take some or all of the individual patient s PRO assessments and aggregate them to summarize the patterns and quality of care received at the clinic or health plan level (Box 3). Examples of this might be for pay-forperformance, or for pubic report cards. The next step in the process is to use the PRO data to inform quality improvement (Box 4). Dr. John Browne from University College Cork has described how this process works using the example of breast reconstruction following mastectomy (Browne 2009). In Browne s example, individual surgeons are presented with the average PRO scores of their reconstruction patients. The surgeons performance falls into a distribution, with some surgeons patients reporting lower PRO scores and other surgeons patients reporting higher PRO scores, on average. While PRO measures provide some indication of relative performance, scores alone are not particularly illuminating in terms of how to improve care. The key is to be able to translate the scores into descriptive 12

14 labels that can inform patient management. In Browne s example, lower scores reflect women who find their breasts shape to be acceptable when clothed; the average scores reflect women reporting that their breasts line up when unclothed; and the highest scores represent women who report that their breasts are equal in size and shape when unclothed. These descriptive labels can inform a lower-performing surgeon regarding what areas require increased attention, making the PROs a powerful tool for quality improvement. The cycle then begins again with PRO assessment (Box 1) ideally, with the PRO scores demonstrating the improvements based on the applications of the PRO data for quality assessment and improvement (Boxes 2-4). In addition, having accumulated a number of PRO reports creates another opportunity for using the PROs in clinical practice (Box 2). That is, group-level PRO data can be assembled in the form of decision aids that can be used to explain the implications of treatment alternatives and help patients and clinicians decide on the appropriate strategy for a given patient. In contrast to the use of an individual s PROs informing his or her care, in this application, PRO data from other patients are summarized and presented to a patient to help him/her understand other patients experiences with the various treatment options. For example, Brundage et al. have shown that lung cancer patients presented with the hypothetical option of chemotherapy used information regarding the impact of chemotherapy on survival, toxicity as well as on health-related quality of life to inform their decision (Brundage 2005). Thus, as more data are collected about patient experiences using PROs, more patients can benefit from a clearer understanding of the quality-of-life implications of different treatment options. This cycle is exemplary of the functioning of a learning health care system one in which best practices are embedded in the delivery process and new knowledge is captured as an integral byproduct of the delivery experience (Olsen 2007). While the above discussion of the applications for PROs in quality measurement and improvement around the cycle focuses primarily on the PROs themselves, the power of PROs increases substantially when the PRO data are linked with other clinical information (Wu 1997, Snyder 2013b). Linkage to treatments and clinical events can provide information about treatment effectiveness for individual patients, while linkage to 13

15 patient and disease and provider characteristics can help to generate evidence about the effectiveness of the care delivered by providers. The integration of PROs in EHRs offers great potential in terms of applying PROs during each phase of the cycle. Below, we provide additional background on EHRs and then describe how PROs, combined with the EHR data, create valuable opportunities for patient-centered care and research. Electronic Health Records (EHRs) A more detailed review of EHRs is beyond the scope of this paper. The following is a brief history of the development of the electronic health record and a description of its current state. An EHR can be defined as a systematic collection of electronic health information in digital format about individual patients or populations. Data may be captured in many ways, including as structured data that can be subjected to immediate analysis and other ways such as free text or images. Information is stored so that it can be accessed across different health care settings including hospitals, clinics and other care facilities, and even by individual patients. EHRs can collect a broad range of data, including patient demographics, medical and social history, medication and allergies, diagnoses and problems, immunization status, laboratory and other test results, vital signs, physical examination findings, billing information and various documents. Additional functions include the ability to execute orders for tests and medications, schedule future appointments, generate referrals to other providers, track care and outcomes, trigger warnings and file public health reports. Information within an EHR can be used for secondary analyses for research, quality assessment, quality improvement and reporting (Weiner 2012). The acronyms EHR and EMR (electronic medical record) are often used interchangeably. An EMR denotes an electronic version of the patient record created for hospitals and clinics whereas an EHR has a broader purpose of giving access to a patient s medical information to multiple stakeholders across different facilities within and institution or network, including patients, health care providers, managers, payers, insurers, and employers. The modern medical record dates back to the 1970s when records were still maintained exclusively on paper (Weed 1972). There were a few initial attempts to digitize these 14

16 records and provide computerized decision support (NIH 2006). Electronic medical records began to be developed in the 1980s for administrative purposes. A notable effort was the Veterans Administration s VisTA which was adopted universally across VA Medical Centers (Brown 2003). In the 1990s the first Windows-based medical records were released (NIH 2006). The scope of EMRs broadened in the early 2000s to include a range of non-clinical health information, leading to the term "electronic health record". Some of these EHRs have integrated patient portals and allowed patients to communicate securely with health providers and to enter additional information. The process was greatly accelerated by passage of the Health Information Technology for Economic and Clinical Health Act and availability of stimulus funds to reward the meaningful use of data by adopters of EHR (ONC 2012). A move to develop standalone personal health records was also initiated (Tang 2006), which is proceeding with increasing success but will not be reviewed here. As these systems have developed, EHR-based patient portals (referred to as tethered portals) have emerged as the most prevalent structure. These portals permit patients to retrieve their records and to enter additional information (Tang 2006). It is uncertain what proportion of EHRs have the capacity to collect PROs. A number of standalone web tools have been designed specifically to capture and report PRO measures, in part in response to the lack of availability of this capability within existing EHRs. Most of these do not interface directly with EHRs. Other example as mentioned above is PatientViewpoint, developed by our group (which does link to the EHR), with other examples of integrated systems identified in the case studies section (Snyder 2009, Snyder 2012). EHRs systems have the potential to enhance patient-clinician interactions through the incorporation of patient-level PRO scores. In some EHRs, clinicians can access reference materials to support decision-making. Other systems have automated alerts and decision support that is built-in to help guide practice precisely when it is needed. Today, over 57% of office-based physicians use EHRs (Hsiao 2011). Concurrently, despite worries about a potential digital divide exacerbating existing disparities in care, internet use is increasingly prevalent for both genders and all age groups, races/ethnicities and income levels. This advance is due in part to internet access from 15

17 laptops, tablets, telephones and other handheld devices (Zickuhr 2012). We are now in the midst of a period of rapid development of EHRs capabilities alongside an even more rapid proliferation of web-based options for PRO collection. A natural place for these data to go is into EHRs, but there is relatively little published on the best practices and new developments in linking PRO data collection to the EHR. For that reason, we undertook a search of the published and grey literature on the current state of use of PRO measures in electronic health records. We were interested in identifying highly developed and innovative systems for PRO data collection that feature EHR integration. We were also interested in common themes in the implementation and integration of PROs within the EHR, including barriers and facilitating factors, and in identifying key unanswered questions important to the future on PRO measure integration in the EHR. METHODS We conducted a landscape (non-systematic) review to identify existing and leading systems that collect PROs and link to EHRs. We were interested in systems that were used in clinical practice. Systems were eligible if they are used in clinical care settings, assess PROs electronically and provide summaries of the patients response to providers. Systems were identified through publications, as well as conference abstracts, white papers, reports published online or in print and other grey literature. The latter included information from unpublished presentations, publications and news reports. We obtained additional detail through interviews with key informants identified by the Planning Committee and our own network of colleagues and collaborators. PubMed, MEDLINE and Embase searches used the following terms: [patient-reported outcomes (outcome assessment, quality of life, health status indicators, patient-reported), and clinical care (patient care, clinical care, delivery of health care)]. We excluded PRO data collection systems intended exclusively for clinical trials. We also excluded electronic PRO data collection systems that were not linked to EHRs. This exclusion category included some major PRO reporting efforts (Meyer 1994, Gustafson 2001, Wasson 1999, Cohen 2013), but allowed this report to 16

18 specifically focus on how PRO data is integrated with EHR information. In addition, we did not include the large number of efforts to measure patient satisfaction independent of PRO measures of health status. We did not include the many other different kinds of health related information, including the review of systems, family history, health behaviors such as tobacco or alcohol use, and health care utilization. We limited our selection of case studies to US systems (Table 1) although there are important international efforts of note (Gilbert 2012, Dudgeon 2012, Black 2013, Varagunam 2013; Bainbridge 2011, Engelen 2010). The cases we present do not represent the full range of systems. While PRO collection and integration are most common in specific clinical populations such as cancer, rheumatology and orthopedics, we selected systems that represent a range of applications from the general population to specific disease conditions. We included cases used in health systems with different payment and organizational models and different scale from small to large. For each case study, system characteristics and clinical implementation were identified and abstracted using a structured review form created by the authors. We supplemented the reviews with interviews with the principal system developers and users. The summary extract of information was verified with the system developers from whom we also obtained follow-up information when necessary. Table 1 summarizes the 11 case studies, including their system affiliation and name, the initial clinical population, and whether they are used at multiple practice locations and for multiple patient populations. 17

19 TABLE 1: SUMMARY OF CASE STUDIES # System Affiliation (Name) Initial Population Multiple Sites/Clinics Multiple Populations 1 Epic Systems Corporation Epic Users Y Y (MyChart, EpicCare) 3 Cleveland Clinic (Knowledge Neurological Y Y Program) Disorders 2 Dartmouth Spine Center Spine Y Y 4 Group Health Cooperative General Y N (Health Profile e-hra) 5 Cincinnati Children s Hospital Rheumatology Y Y 6 Kaiser Permanante Colorado Older Adults Y N (PATHWAAY) 7 Essentia Health (MN Depression Y N Community Measurement) 8 University of Pittsburgh Medical Primary Care Y Y Center 9 Duke University (Patient Care Cancer Y Y Monitor) 10 UCLA/Michigan (My GI-Health) GI Disorders Y N 11 University of Washington/ Centers for AIDS Research Networks of Clinical Systems HIV Y N For each case, the first page provides an overview of the system and its development, including conditions and PROs included; the nature of integration of PRO measures within the EHR, applications in clinical practice, research and quality improvement, and future plans for the system. The facing page shows a graphic example of one aspect of the system, a walk-through of the process of patient assessment and data flow, key themes highlighted by the case, and data sources. 18

20 Epic Systems Corporation MyChart Basic System Summary: MyChart (Epic Systems Corporation, Verona, WI) is a secure member website through which registered patients can view portions of their medical record and exchange secure messages with physicians. Although collection of PROs within Epic had been implemented prior to the 2012 release of MyChart, the release of the series feature for PRO ordering and definition tool added features that added value for researchers. Epic has been granted permission to provide the following PRO measures as part of their Foundation System. They are: Measure Medical Outcomes Study PHQ2/PHQ9 PROMIS Adult (18 years and greater) static short forms Pediatric Self Report (8-17 years) static short forms Proxy (5-17 years) static short forms Dimensions SF-20 (20 item Short Form health survey), RAND-36 Depression Physical Functioning (10 items); Pain Interference (8 items); Global Rating of Pain (1 item); Sleep Disturbance (8 items); Fatigue (8 items); Depression (8 items); Anxiety (8 items); Satisfaction with Participation in Social Roles (8 items) Physical Functioning--Mobility (8 items); Pain Interference (8 items); Global Rating of Pain (1 item); Fatigue (10 items); Depressive Symptoms (8 items); Anxiety (8 items); Peer Relationships (8 items) Physical Functioning--Mobility (8 items); Pain Interference (8 items); Global Rating of Pain (1 item); Fatigue (10 items); Depressive Symptoms (8 items); Anxiety (8 items); Peer Relationships (8 items) EHR Integration: Complete integration with the Epic EHR. PRO scores can be viewed and manipulated alongside other clinical data elements such as laboratory test results. Clinical Practice: Epic series definition functionality makes it possible on an individual or sub-population basis (e.g., all patients age > 65) to specify the timings and intervals of automated releases of one or more PRO measures. Organizations can build additional EpicCare/MyChart questionnaires Research-Related: It is possible to specify the timing and intervals of automated releases of PRO assessments. PRO data are aggregated across patients in Epic s analytic environments (Cogito Clarity and Cogito Data Warehouse) with common data structures across organizations. Quality Improvement: Organizations that have implemented Clarity and/or the data warehouse are able to develop reports that include PRO data. Future Plans: Long-term plans are to expand to additional PROs in their Foundation system. 19